[PATCH 0/3] ext4 freeze regression test

[PATCH 0/3] ext4 freeze regression test

[Amir Goldstein]

Eryu,

This is a regression test for a fix waiting in Ted's dev branch
preceded by two minor cleanups.

FYI, at the moment, all _require_freeze tests fail with lockdep enabled
because of DEBUG_RWSEMS:
https://marc.info/?l=linux-fsdevel&m=152622016219975&w=2

Thanks,
Amir.

Amir Goldstein (3):
  generic/459: use xfs_freeze instead of fsfreeze
  fstests: define and use TIMEOUT_PROG
  generic: test first read with freeze right after mount

 common/config         |  1 +
 tests/generic/297     |  4 +--
 tests/generic/298     |  4 +--
 tests/generic/459     |  6 ++---
 tests/generic/488     | 75 +++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/generic/488.out |  2 ++
 tests/generic/group   |  1 +
 7 files changed, 86 insertions(+), 7 deletions(-)
 create mode 100755 tests/generic/488
 create mode 100644 tests/generic/488.out

-- 
2.7.4

[PATCH 1/3] generic/459: use xfs_freeze instead of fsfreeze

[Amir Goldstein]

This is what all other tests use as well as _require_freeze

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 tests/generic/459 | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tests/generic/459 b/tests/generic/459
index e755cf7..22b8e62 100755
--- a/tests/generic/459
+++ b/tests/generic/459
@@ -113,7 +113,7 @@ $XFS_IO_PROG -f -d -c 'pwrite -b 1m 0 120m' $SCRATCH_MNT/f1 >>$seqres.full 2>&1
 # extended. It is expected, and is only used so xfsaild is triggered to
 # flush AIL items, other filesystems usually get remounted as read-only during
 # the above write process.
-fsfreeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
+xfs_freeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
 freezeid=$!
 
 # Wait enough so xfsaild can run
@@ -132,7 +132,7 @@ ret=$?
 #	- The filesystem stays in Read-Write mode, but can be frozen/thawed
 #	  without getting stuck.
 if [ $ret -ne 0 ]; then
-	# fsfreeze failed, filesystem should reject further writes and remount
+	# freeze failed, filesystem should reject further writes and remount
 	# as readonly. Sometimes the previous write process won't trigger
 	# ro-remount, e.g. on ext3/4, do additional touch here to make sure
 	# filesystems see the metadata I/O error.
@@ -148,7 +148,7 @@ if [ $ret -ne 0 ]; then
 else
 	# Try to thaw the filesystem, and complete test if if succeed.
 	# NOTE: This will hang on affected XFS filesystems.
-	fsfreeze -u $SCRATCH_MNT >>$seqres.full 2>&1
+	xfs_freeze -u $SCRATCH_MNT >>$seqres.full 2>&1
 	echo "Test OK"
 fi
 
-- 
2.7.4

Re: [PATCH 1/3] generic/459: use xfs_freeze instead of fsfreeze

[Dave Chinner]

On Mon, May 14, 2018 at 08:49:49AM +0300, Amir Goldstein wrote:
> This is what all other tests use as well as _require_freeze
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> ---
>  tests/generic/459 | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/tests/generic/459 b/tests/generic/459
> index e755cf7..22b8e62 100755
> --- a/tests/generic/459
> +++ b/tests/generic/459
> @@ -113,7 +113,7 @@ $XFS_IO_PROG -f -d -c 'pwrite -b 1m 0 120m' $SCRATCH_MNT/f1 >>$seqres.full 2>&1
>  # extended. It is expected, and is only used so xfsaild is triggered to
>  # flush AIL items, other filesystems usually get remounted as read-only during
>  # the above write process.
> -fsfreeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
> +xfs_freeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
>  freezeid=$!
>  
>  # Wait enough so xfsaild can run
> @@ -132,7 +132,7 @@ ret=$?
>  #	- The filesystem stays in Read-Write mode, but can be frozen/thawed
>  #	  without getting stuck.
>  if [ $ret -ne 0 ]; then
> -	# fsfreeze failed, filesystem should reject further writes and remount
> +	# freeze failed, filesystem should reject further writes and remount
>  	# as readonly. Sometimes the previous write process won't trigger
>  	# ro-remount, e.g. on ext3/4, do additional touch here to make sure
>  	# filesystems see the metadata I/O error.
> @@ -148,7 +148,7 @@ if [ $ret -ne 0 ]; then
>  else
>  	# Try to thaw the filesystem, and complete test if if succeed.
>  	# NOTE: This will hang on affected XFS filesystems.
> -	fsfreeze -u $SCRATCH_MNT >>$seqres.full 2>&1
> +	xfs_freeze -u $SCRATCH_MNT >>$seqres.full 2>&1
>  	echo "Test OK"
>  fi

Shouldn't these all be converted to XFS_FREEZE_PROG w/ a setup call
to export XFS_FREEZE_PROG="`set_prog_path xfs_freeze`"?

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 1/3] generic/459: use xfs_freeze instead of fsfreeze

[Amir Goldstein]

On Tue, May 15, 2018 at 2:01 AM, Dave Chinner <david@fromorbit.com> wrote:
> On Mon, May 14, 2018 at 08:49:49AM +0300, Amir Goldstein wrote:
>> This is what all other tests use as well as _require_freeze
>>
>> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
>> ---
>>  tests/generic/459 | 6 +++---
>>  1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/tests/generic/459 b/tests/generic/459
>> index e755cf7..22b8e62 100755
>> --- a/tests/generic/459
>> +++ b/tests/generic/459
>> @@ -113,7 +113,7 @@ $XFS_IO_PROG -f -d -c 'pwrite -b 1m 0 120m' $SCRATCH_MNT/f1 >>$seqres.full 2>&1
>>  # extended. It is expected, and is only used so xfsaild is triggered to
>>  # flush AIL items, other filesystems usually get remounted as read-only during
>>  # the above write process.
>> -fsfreeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
>> +xfs_freeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
>>  freezeid=$!
>>
>>  # Wait enough so xfsaild can run
>> @@ -132,7 +132,7 @@ ret=$?
>>  #    - The filesystem stays in Read-Write mode, but can be frozen/thawed
>>  #      without getting stuck.
>>  if [ $ret -ne 0 ]; then
>> -     # fsfreeze failed, filesystem should reject further writes and remount
>> +     # freeze failed, filesystem should reject further writes and remount
>>       # as readonly. Sometimes the previous write process won't trigger
>>       # ro-remount, e.g. on ext3/4, do additional touch here to make sure
>>       # filesystems see the metadata I/O error.
>> @@ -148,7 +148,7 @@ if [ $ret -ne 0 ]; then
>>  else
>>       # Try to thaw the filesystem, and complete test if if succeed.
>>       # NOTE: This will hang on affected XFS filesystems.
>> -     fsfreeze -u $SCRATCH_MNT >>$seqres.full 2>&1
>> +     xfs_freeze -u $SCRATCH_MNT >>$seqres.full 2>&1
>>       echo "Test OK"
>>  fi
>
> Shouldn't these all be converted to XFS_FREEZE_PROG w/ a setup call
> to export XFS_FREEZE_PROG="`set_prog_path xfs_freeze`"?
>

Yeh, I suppose I can take this one as part of the "freeze" cleanup.

Thanks,
Amir.

Re: [PATCH 1/3] generic/459: use xfs_freeze instead of fsfreeze

[Amir Goldstein]

On Tue, May 15, 2018 at 8:25 AM, Amir Goldstein <amir73il@gmail.com> wrote:
> On Tue, May 15, 2018 at 2:01 AM, Dave Chinner <david@fromorbit.com> wrote:
>> On Mon, May 14, 2018 at 08:49:49AM +0300, Amir Goldstein wrote:
>>> This is what all other tests use as well as _require_freeze
>>>
>>> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
>>> ---
[...]
>>
>> Shouldn't these all be converted to XFS_FREEZE_PROG w/ a setup call
>> to export XFS_FREEZE_PROG="`set_prog_path xfs_freeze`"?
>>
>
> Yeh, I suppose I can take this one as part of the "freeze" cleanup.
>

I looked again and xfs_freeze pattern repeats itself quite often with
other xfs_* progs, so I guess I'll leave that cleanup for another time...

Thanks,
Amir.

$ git grep -v '^[[:space:]]*[#;]' tests/*/???| grep -v echo > all
$ (cd /usr/sbin; ls xfs_*) > xfsprogs
$ for p in `cat xfsprogs`; do echo $p; grep -w $p all | wc -l; done
xfs_admin
1
xfs_bmap
24
xfs_copy
9
xfs_db
35
xfs_estimate
0
xfs_freeze
17
xfs_fsr
15
xfs_growfs
3
xfs_info
31
xfs_io
17
xfs_logprint
0
xfs_mdrestore
7
xfs_metadump
1
xfs_mkfile
1
xfs_ncheck
0
xfs_quota
43
xfs_rtcp
0
xfs_spaceman
0

Re: [PATCH 1/3] generic/459: use xfs_freeze instead of fsfreeze

[Dave Chinner]

On Tue, May 15, 2018 at 04:50:04PM +0300, Amir Goldstein wrote:
> On Tue, May 15, 2018 at 8:25 AM, Amir Goldstein <amir73il@gmail.com> wrote:
> > On Tue, May 15, 2018 at 2:01 AM, Dave Chinner <david@fromorbit.com> wrote:
> >> On Mon, May 14, 2018 at 08:49:49AM +0300, Amir Goldstein wrote:
> >>> This is what all other tests use as well as _require_freeze
> >>>
> >>> Signed-off-by: Amir Goldstein <amir73il@gmail.com>
> >>> ---
> [...]
> >>
> >> Shouldn't these all be converted to XFS_FREEZE_PROG w/ a setup call
> >> to export XFS_FREEZE_PROG="`set_prog_path xfs_freeze`"?
> >>
> >
> > Yeh, I suppose I can take this one as part of the "freeze" cleanup.
> >
> 
> I looked again and xfs_freeze pattern repeats itself quite often with
> other xfs_* progs, so I guess I'll leave that cleanup for another time...

No worries, just add it to the pile...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 1/3] generic/459: use xfs_freeze instead of fsfreeze

[Eryu Guan]

On Mon, May 14, 2018 at 08:49:49AM +0300, Amir Goldstein wrote:
> This is what all other tests use as well as _require_freeze
> 
> Signed-off-by: Amir Goldstein <amir73il@gmail.com>

I'll drop this patch for now per discussion but take the other two
patches, the XFS_FREEZE_PROG conversion is not urgent and could wait I
think :)

Thanks,
Eryu

> ---
>  tests/generic/459 | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/tests/generic/459 b/tests/generic/459
> index e755cf7..22b8e62 100755
> --- a/tests/generic/459
> +++ b/tests/generic/459
> @@ -113,7 +113,7 @@ $XFS_IO_PROG -f -d -c 'pwrite -b 1m 0 120m' $SCRATCH_MNT/f1 >>$seqres.full 2>&1
>  # extended. It is expected, and is only used so xfsaild is triggered to
>  # flush AIL items, other filesystems usually get remounted as read-only during
>  # the above write process.
> -fsfreeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
> +xfs_freeze -f $SCRATCH_MNT >>$seqres.full 2>&1 &
>  freezeid=$!
>  
>  # Wait enough so xfsaild can run
> @@ -132,7 +132,7 @@ ret=$?
>  #	- The filesystem stays in Read-Write mode, but can be frozen/thawed
>  #	  without getting stuck.
>  if [ $ret -ne 0 ]; then
> -	# fsfreeze failed, filesystem should reject further writes and remount
> +	# freeze failed, filesystem should reject further writes and remount
>  	# as readonly. Sometimes the previous write process won't trigger
>  	# ro-remount, e.g. on ext3/4, do additional touch here to make sure
>  	# filesystems see the metadata I/O error.
> @@ -148,7 +148,7 @@ if [ $ret -ne 0 ]; then
>  else
>  	# Try to thaw the filesystem, and complete test if if succeed.
>  	# NOTE: This will hang on affected XFS filesystems.
> -	fsfreeze -u $SCRATCH_MNT >>$seqres.full 2>&1
> +	xfs_freeze -u $SCRATCH_MNT >>$seqres.full 2>&1
>  	echo "Test OK"
>  fi
>  
> -- 
> 2.7.4
> 

[PATCH 2/3] fstests: define and use TIMEOUT_PROG

[Amir Goldstein]

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 common/config     | 1 +
 tests/generic/297 | 4 ++--
 tests/generic/298 | 4 ++--
 3 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/common/config b/common/config
index cc31806..af360ce 100644
--- a/common/config
+++ b/common/config
@@ -204,6 +204,7 @@ export UBIUPDATEVOL_PROG="`set_prog_path ubiupdatevol`"
 export THIN_CHECK_PROG="$(set_prog_path thin_check)"
 export PYTHON2_PROG="`set_prog_path python2`"
 export SQLITE3_PROG="`set_prog_path sqlite3`"
+export TIMEOUT_PROG="`set_prog_path timeout`"
 
 # use 'udevadm settle' or 'udevsettle' to wait for lv to be settled.
 # newer systems have udevadm command but older systems like RHEL5 don't.
diff --git a/tests/generic/297 b/tests/generic/297
index f7eb37b..a3ba5c4 100755
--- a/tests/generic/297
+++ b/tests/generic/297
@@ -45,7 +45,7 @@ _cleanup()
 _supported_os Linux
 _require_scratch_reflink
 _require_cp_reflink
-_require_command "$(which timeout)" "timeout"
+_require_command "$TIMEOUT_PROG" "timeout"
 
 test $FSTYP == "nfs"  && _notrun "NFS can't interrupt clone operations"
 
@@ -83,7 +83,7 @@ kill_after=2	# give us a shorter time to die
 n=$(stat -c '%s' $testdir/file1)
 echo "performing kill test on $n bytes..." >> $seqres.full
 touch $TEST_DIR/before
-timeout -s INT ${kill_after}s $XFS_IO_PROG -f -c "reflink $testdir/file1 0 $n $n" $testdir/file1 >> $seqres.full 2>&1
+$TIMEOUT_PROG -s INT ${kill_after}s $XFS_IO_PROG -f -c "reflink $testdir/file1 0 $n $n" $testdir/file1 >> $seqres.full 2>&1
 touch $TEST_DIR/after
 before=$(stat -c '%Y' $TEST_DIR/before)
 after=$(stat -c '%Y' $TEST_DIR/after)
diff --git a/tests/generic/298 b/tests/generic/298
index b518da1..a4d458c 100755
--- a/tests/generic/298
+++ b/tests/generic/298
@@ -45,7 +45,7 @@ _cleanup()
 _supported_os Linux
 _require_scratch_reflink
 _require_cp_reflink
-_require_command "$(which timeout)" "timeout"
+_require_command "$TIMEOUT_PROG" "timeout"
 
 test $FSTYP == "nfs"  && _notrun "NFS can't interrupt clone operations"
 
@@ -83,7 +83,7 @@ kill_after=2	# give us a shorter time to die
 n=$(stat -c '%s' $testdir/file1)
 echo "performing kill test on $n bytes..." >> $seqres.full
 touch $TEST_DIR/before
-urk=$(timeout -s KILL ${kill_after}s $XFS_IO_PROG -f -c "reflink $testdir/file1 0 $n $n" $testdir/file1 >> $seqres.full 2>&1)
+urk=$($TIMEOUT_PROG -s KILL ${kill_after}s $XFS_IO_PROG -f -c "reflink $testdir/file1 0 $n $n" $testdir/file1 >> $seqres.full 2>&1)
 touch $TEST_DIR/after
 before=$(stat -c '%Y' $TEST_DIR/before)
 after=$(stat -c '%Y' $TEST_DIR/after)
-- 
2.7.4

[PATCH 3/3] generic: test first read with freeze right after mount

[Amir Goldstein]

With ext4, this leads to freeze proection bypass WARN_ON in
ext4_journal_check_start.

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
---
 tests/generic/488     | 75 +++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/generic/488.out |  2 ++
 tests/generic/group   |  1 +
 3 files changed, 78 insertions(+)
 create mode 100755 tests/generic/488
 create mode 100644 tests/generic/488.out

diff --git a/tests/generic/488 b/tests/generic/488
new file mode 100755
index 0000000..02cf489
--- /dev/null
+++ b/tests/generic/488
@@ -0,0 +1,75 @@
+#! /bin/bash
+# FS QA Test No. 488
+#
+# Test first read with freeze right after mount.
+#
+# With ext4, this leads to freeze proection bypass WARN_ON in
+# ext4_journal_check_start.
+#
+#-----------------------------------------------------------------------
+# Copyright (C) 2018 CTERA Networks. All Rights Reserved.
+# Author: Amir Goldstein <amir73il@gmail.com>
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+#-----------------------------------------------------------------------
+#
+
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	cd /
+	rm -f $tmp.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# real QA test starts here
+_supported_fs generic
+_supported_os Linux
+_require_scratch
+_require_freeze
+_require_command "$TIMEOUT_PROG" "timeout"
+
+rm -f $seqres.full
+
+testfile=$SCRATCH_MNT/foo
+
+# Create a filesystem with a file
+_scratch_mkfs >>$seqres.full 2>&1
+_scratch_mount
+echo "frozen" > $testfile
+
+# Mount filesystem "noatime" and freeze right after mount -
+# reads should not block
+_scratch_cycle_mount "noatime"
+xfs_freeze -f $SCRATCH_MNT
+
+# Read file while filesystem is frozen should succeed
+# without blocking
+$TIMEOUT_PROG -s KILL 1s cat $testfile
+
+xfs_freeze -u $SCRATCH_MNT
+
+status=0
+exit
diff --git a/tests/generic/488.out b/tests/generic/488.out
new file mode 100644
index 0000000..0907d7c
--- /dev/null
+++ b/tests/generic/488.out
@@ -0,0 +1,2 @@
+QA output created by 488
+frozen
diff --git a/tests/generic/group b/tests/generic/group
index dc637c9..ec993f7 100644
--- a/tests/generic/group
+++ b/tests/generic/group
@@ -490,3 +490,4 @@
 485 auto quick insert
 486 auto quick attr
 487 auto quick
+488 auto quick freeze mount
-- 
2.7.4